Have we considered arranging it so that a large language model is struck by lightning?

If there's one thing I know, it's that lightning = sentience.